Molecular characterization of the family of choline transporter-like proteins and their splice variants

Authors

  • E. Traiffort,

    1. Institut de Neurobiologie Alfred Fessard IFR 2118 CNRS, Laboratoire de Neurobiologie Cellulaire et Moléculaire, UPR 9040, Gif-sur-Yvette, France
    Search for more papers by this author
  • M. Ruat,

    1. Institut de Neurobiologie Alfred Fessard IFR 2118 CNRS, Laboratoire de Neurobiologie Cellulaire et Moléculaire, UPR 9040, Gif-sur-Yvette, France
    Search for more papers by this author
  • S. O'Regan,

    1. Institut de Neurobiologie Alfred Fessard IFR 2118 CNRS, Laboratoire de Neurobiologie Cellulaire et Moléculaire, UPR 9040, Gif-sur-Yvette, France
    Search for more papers by this author
  • F. M. Meunier

    1. Institut de Neurobiologie Alfred Fessard IFR 2118 CNRS, Laboratoire de Neurobiologie Cellulaire et Moléculaire, UPR 9040, Gif-sur-Yvette, France
    Search for more papers by this author

Address correspondence and reprint requests to E. Traiffort, UPR 9040 CNRS, 1 Avenue de la Terrasse, 91198 Gif-sur-Yvette, France.
E-mail: Elisabeth.Traiffort@nbcm.cnrs-gif.fr

Abstract

We show here that the choline transporter-like (CTL) family is more extensive than initially described with five genes in humans and complex alternative splicing. In adult rat tissues, CTL2–4 mRNAs are mainly detected in peripheral tissues, while CTL1 is widely expressed throughout the nervous system. During rat post-natal development, CTL1 is expressed in several subpopulations of neurones and in the white matter, where its spatio-temporal distribution profile recalls that of myelin basic protein, an oligodendrocyte marker. We identified two major rat splice variants of CTL1 (CTL1a and CTL1b) differing in their carboxy-terminal tails with both able to increase choline transport after transfection in neuroblastoma cells. In the developing brain, CTL1a is expressed in both neurones and oligodendroglial cells, whereas CTL1b is restricted to oligodendroglial cells. These findings suggest specific roles for CTL1 splice variants in both neuronal and oligodendrocyte physiology.

Abbreviations used:
CTL

choline transporter-like

EST

Expressed sequence tag

HC-3

hemicholinium-3

ISH

in situ hybridization

MBP

myelin basic protein

P

post-natal day

PBS

phosphate-buffered saline

rCTL1

rat CTL1

SSC

sodium citrate buffer

SSPE

sodium soline phosphate EDTA buffer

tCTL1

torpedo CTL1

Choline has been mainly studied as both a component of membrane phospholipids and as a precursor for the synthesis of the neurotransmitter acetylcholine but choline metabolites also serve as osmolytes and intracellular messengers (Blusztajn 1998). The pharmacological characterization of mechanisms for choline transport has allowed low- and high-affinity choline transports to be distinguished (for review see Lockman and Allen 2002). Initially characterized as a process of facilitated diffusion through the cell membrane (Yamamura and Snyder 1973) and then suggested to be a carrier-mediated uptake (Ferguson et al. 1991), low-affinity choline transport has been detected ubiquitously. On the other hand, sodium-dependent high-affinity choline transport is restricted to cholinergic nerve terminals for the synthesis of acetylcholine and is antagonized by low concentrations of the choline analogue hemicholinium-3 (HC-3) (Yamamura and Snyder 1972; Kuhar and Murrin 1978). However, pharmacological and biochemical studies have yet reported the existence of other choline transport mechanisms that cannot be classified in either of these two groups. For instance, the endothelial cells of the blood–brain barrier display an intermediate affinity for choline and HC-3 (Allen and Smith 2001; Lockman et al. 2001) and extracellular choline is efficiently taken up and incorporated into phospholipids by many types of cultured cells, including neuroblastoma (Lanks et al. 1974), glioma (George et al. 1989) and the hybrid NG108-15 (McGee 1980).

The molecular approach to identifying the mechanisms of choline transport has recently allowed the association of a given protein with a few of the choline uptake systems described above. Thus, the organic cation transporter OCT2 has been proposed as a low-affinity choline transporter across the ventricular membrane of the choroid plexuses (Sweet et al. 2001) and two higher affinity choline transporters belonging to different protein families have been cloned. The first, CHT1, displays all the characteristics of the canonical high-affinity choline transporter associated with cholinergic nerve terminals (Okuda et al. 2000; Misawa et al. 2001). The second, CTL1, belongs to a new family of choline transporter-like (CTL) proteins. CTL1-induced transport has only weak sodium dependence (O'Regan and Meunier 2003) and CTL1 is expressed in several neuronal populations, including motorneurones, as well as in oligodendrocytes (O'Regan et al. 2000).

In the present work, we have determined the tissue distribution of the different CTL family members and then, focusing on CTL1 as the major member of the family expressed in the nervous system, analysed the temporal and spatial regulation of CTL1 transcripts during post-natal development. Moreover, we have identified alternative splicing mechanisms for three of five CTL family genes in mammals. CTL1a and CTL1b splice variants encode two proteins differing only by a few amino acids at their C-termini and transfection of either cDNA leads to a moderate increase in choline uptake in transfected eukaryotic cells. As indicated by their brain distribution, both variants might be implicated in oligodendroglial cell function during development, whereas CTL1a may suffice in neurones.

Experimental procedures

Database analysis

Sequence similarity searches against the available database for expressed sequence tags (ESTs) and genomes were performed using the BLAST program at NCBI (http://www.ncbi.nlm.nih.gov/BLAST/). ESTs were assembled into contigs with the CAP program (http://www.infobiogen.fr/services/analyseq/cgi-bin/cap_in.pl). Alignments of cDNA with genomic sequences were optimized manually according to consensus sequences for intron splicing. Proteins were analysed with Clustal W at Infobiogen (http://www.infobiogen.fr/services/analyseq/cgi-bin/clustalw_in.pl) for the guide tree and alignment.

Northern blot analysis

Total RNA was isolated from adult male Wistar rat brain and peripheral tissues using the extraction solution RNAble (Eurobio, Les Ulis, France) and poly(A)+RNA was selected using oligo(dT)-cellulose affinity chromatography. The northern blots were prepared as described previously (Traiffort et al. 1998). Hybridization was performed overnight at 42°C in a solution containing 50% deionized formamide, 5 × Denhart's solution, 5 × sodium saline phosphate EDTA buffer (SSPE), 0.2% sodium dodecyl sulphate, 0.75 mg/mL heat-denaturated salmon sperm DNA and [α-32P]dCTP-labelled DNA probes. The blots were firstly washed twice at 42°C in 2 × saline sodium citrate buffer (SSC) and 0.1% sodium dodecyl sulphate and secondly at 50°C in 0.2 × SSC and 0.1% sodium dodecyl sulphate. Finally, the blots were set up with an autoradiography film and exposed with amplifying screens at −80°C.

DNA and RNA probe synthesis

DNA fragments of CTL1, CTL1a and CTL1b were obtained by restriction or PCR amplification of clones isolated from a rat cDNA library. They corresponded to nucleotides 19–1651 for CTL1 (GenBank Accession no. AJ245619) and to 1412 or 908 bp of the non-coding sequence adjacent to the stop codon of CTL1a (GenBank Accession no. AJ420809) and CTL1b (GenBank Accession no. AJ245619), respectively. Mouse CTL2 probe was obtained by PCR amplification of cDNA from mouse neuroblastoma N18 cells corresponding to nucleotides 907–2029 from GenBank Accession no. AK048648.1. Mouse CTL3 DNA probe was prepared using an XhoI/EcoRV restriction fragment from the Image clone (GenBank Accession no. W64177), corresponding to nucleotides 1402–1696 of GenBank Accession no. BC010552. Mouse CTL4 probe corresponded to nucleotides 3–2236 of GenBank Accession no. AA245579. In all cases, 32P-labelled DNA probes were prepared by random priming from the appropriate matrix using the Nonaprimer Kit (Appligene, Illkirch Graffenstaden, France). Antisense or sense 11-UTP digoxigenin or 12-UTP fluorescein riboprobes were transcribed using T7 or SP6 RNA polymerase (Roche Molecular Biochemicals, Meylan, France). Rat choline acetyltransferase and myelin basic protein (MBP) cRNA probes were as previously described (Traiffort et al. 1999; Ferry et al. 2000).

In situ hybridization

Post-natal or adult male Wistar rats were killed by decapitation, the brain was rapidly removed from the cranium, frozen in cold isopentane maintained in liquid N2 and immediately sectioned. The fifth lumbar dorsal root ganglia were removed from adult rats and treated as described for brain. Cryostat sections (20 µm) were prepared on superfrost Plus/slides (Menzel-Gläser, Braunschweig, Germany) and stored at −80°C until use. In situ hybridization (ISH) experiments were adapted from Schaeren-Wiemers and Gerfin-Moser (1993). Tissue sections were fixed by immersion in freshly made 4% paraformaldehyde in 1 × phosphate-buffered saline (PBS), pH 7.4, rinsed three times in 1 × PBS, acetylated for 10 min at 20°C in a mixture containing 1.33% triethanolamine, 0.2% HCl 10 n and 0.25% acetic anhydride in water and rinsed three times in 1 × PBS. Slides were pre-hybridized for 2 h at room temperature with 400 µL hybridization solution (50% formamide, 5 × SSC, 5 × Denhardt's solution, 500 µg/mL herring sperm DNA). Hybridization was performed overnight at 72°C, under glass siliconized coverslips, in 150 µL of the previously described hybridization solution containing 0.5 µg/mL of digoxigenin-labelled riboprobes. After removal of coverslips, slides were successively washed twice for 45 min at 72°C in 0.2 × SSC, incubated for 30 min at 37°C in the presence of 6 µg/mL ribonuclease A, washed for 2 × 30 min in 0.2 × SSC at the same temperature and finally incubated for 1 h at room temperature in 0.1 m Tris-HCl, pH 7.5, 0.15 m NaCl and 1% heat inactivated normal goat serum. Immunological detection was performed by overnight incubation at 4°C in the last buffer supplemented with a 1/5000 dilution of a sheep anti-digoxigenin antibody conjugated to alkaline phosphatase and revealed with nitroblue tetrazolium and 5-bromo-4-chloro-3-indolyl-phosphate as substrates in the presence of 0.24 mg/mL levamisole. For double ISH experiments, hybridization was performed in the previously described solution containing 0.5 µg/mL of both digoxigenin- and fluorescein-labelled riboprobes. The immunological detection used first a 1/1000 dilution of a sheep anti-fluorescein antibody conjugated to alkaline phosphatase (Roche Molecular Biochemicals) followed by a colorimetric detection with Fast Red as a substrate. Pictures of stained cells were obtained before immunodetection of the second labelled riboprobe using the sheep anti-digoxigenin antibody conjugated to alkaline phosphatase as previously mentioned. Before mounting the slides, the first staining was removed using ethanol and xylene baths. Cell counting was performed on the pictures of cells stained with Fast Red and nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl-phosphate, respectively, and derived from three to four consecutive slices of two different animals.

Care was taken to minimize animal suffering and all procedures were carried out in accordance with the European Communities Council Directive (86/609/EEC) and the French guidelines for the care and use of laboratory animals.

cDNA isolation of CTL1a splice variant

Full-length cDNAs for CTL1a and CTL1b were obtained from a rat brain library constructed in a bacteriophage λ vector (ZAPII- cDNA synthesis kit; Stratagene, La Jolla, CA, USA) using a PCR probe based on rat sequences homologous to the original Torpedo marmorata CTL1 sequence. They were first distinguished on the basis of different restriction patterns. The clones were excised and fully sequenced. The final sequences were deposited in GenBank with the Accession nos AJ420809 and AJ245619.

N18 cell transfection and [3H]choline uptake

N18 cells were grown to 70% confluency on 100-mm dishes, transfected with 5 µg of pcDNA3/Neo (Invitrogen, Cergy Pontoise, France) either as such or with cDNA inserts for rat CTL1a or CTL1b using FuGENE 6 transfection reagent (Roche Molecular Biochemicals). The medium was changed the next morning and experiments carried out 2 h later. The cells were harvested in 15 mL of warm PBS containing 1.5 mm CaCl2, 1.3 mm MgSO4 and 1.8 g/L glucose sent as a medium pressure stream with an automatic pipette and collected by centrifugation at 1000 g for 5 min. The cells were resuspended in 1.5 mL warm supplemented PBS, aliquoted at 100 µL per 1.5-mL microtube and warmed to 37°C in a water-bath for 5 min during which unlabelled choline or HC-3 (Sigma, Saint Quentin Fallavier, France) was added at the indicated concentrations. Choline uptake was measured by adding 0.1 µCi of [3H]choline (75 Ci/mmol; Amersham Biosciences, Orsay, France), followed 6 min later by the addition of 750 µL of cold PBS to stop uptake. The cells were centrifuged at 1000 g for 5 min in a benchtop centrifuge, then resuspended in 0.5% Triton X-100 and assayed for accumulated radioactivity and protein content. Uptake data were corrected for background levels ascertained in the presence of 1 mm unlabelled choline and presented either as rates of transport based on the specific activity of choline in the medium for the saturation analysis or as percentage of choline transport remaining when different concentrations of the inhibitor, HC-3, were added compared with the levels of uptake observed in each case in the absence of the drug. Km values were obtained by fitting the data with a one site binding hyperbola using GraphPad Prism software (GraphPad Prism Software Inc., San Diego, CA, USA). Data are given as means ± SEM of four independent experiments.

Results

Molecular characterization of choline transporter-like family members and identification of choline transporter-like splice variants

Homology searching in databases of conceptual translations of genomic sequences and of expressed sequence tags allowed a more extensive analysis of the CTL gene family (Fig. 1). As reported previously (O'Regan et al. 2000), CTL orthologues are found throughout the phylogenetic tree of all eukaryotes, while no prokaryote homologue has been found. A single CTL gene has been identified in yeast (ScCTL, YOR161c) and Caenorhabditis elegans (CeCTL, F35C8.7), while insects have either two homologues as in the fruit fly (DmCTL1, CG1311 and DmCTL2, CG11880) or three homologues as in mosquito (agCP5661, agCP10678 and agCP12603, data not shown). One CTL homologue (AtCTL, At3G15380) is found in plants (Fig. 1a) and five other genes are present in the Arabidopsis genome which are closer to the yeast gene ScCTL (data not shown). In addition to the four genes previously reported in mammals, we have now identified, by database screening, a fifth gene (CTL5) that appeared recently in the human and macaque genome databases. The small number of human expressed sequence tags (42 compared with 660 for hCTL2 or 480 for hCTL1) indicates that CTL5 is much less expressed than the other homologues. The CTL5 protein is closer to the CTL2/CTL4 than to the CTL1/CTL3 subgroup (Figs 1a and c). The diversity and complexity of the CTL family is further increased by alternative splicing mechanisms. The genomic organization of human CTL1 was also deduced from database analysis. The gene displays 17 exons spanning about 200 kb and contains an alternative splicing site located at the 3′ end of exon 15 leading to transcripts CTL1a (use of exon 16), CTL1b (exon 16 skipped, use of exon 17) and CTL1c showing an internal splice site in exon 16 (Wille et al. 2001) (Figs 1b and c). An identical genomic organization gives rise to the CTL1a and CTL1b transcripts in the mouse and rat. Likewise, the genomic sequences for human CTL2 and CTL5 contain the corresponding exons from which the alternatively spliced transcripts can be generated, separated by up to 20–50 kb of intronic sequence (data not shown).

Figure 1.

Phylogenetic analysis of choline transporter-like (CTL) family members and gene organization of CTL1 splice variants. (a) Dendrogram built using Clustal W program with the mammalian CTL proteins from human [hCTL1a,b,c (AJ420812, AJ245620 and AJ272365), hCTL2a,b (AJ245621 and AL832517), hCTL3 (AX524865), hCTL4/ng22 (AAD21813), hCTL5a,b (BC034580 and BC028743)], rat [rCTL1a,b (AJ420809 and AJ245619)], mouse [mCTL2a,b (AAH31535 and AA647643), mCTL3 (AAH10552), mCTL4/ng22 (AAC84166)] and macaque [MfCTL5 (AB070066)] and with conceptual proteins from the genome of Saccharomyces cerevisiae (ScCTL, YOR161c), Arabidopsis (AtCTL, At3G15380), Drosophila melanogaster (DmCTL1, CG1311; DmCTL2, CG11880) and C. elegans (CeCTL, F35C8.7b). (b) Scheme of the last exons of the human CTL1 gene (contig NT_008470.15). Shaded boxes show the exons making up the three reported splice isoforms and white boxes indicate the spliced exons. The position of the stop codon in the coding sequence (Stop) is indicated by a dark dot. (c) Alignment of the C-terminal part of the rat proteins CTL1a and CTL1b with their human homologues, encoded by the indicated exons. The arrows in the upper line represent the exon limits (for CTL1 and CTL3 only). The numbers flanking the sequences indicate the position of the last amino acid.

From a rat brain cDNA library, we isolated two different CTL1 orthologues, CTL1a (656 amino acid residues; present work) and CTL1b (653 amino acid residues; O'Regan et al. 2000) generated by alternative splicing of their C-terminal end. We sequenced two cDNAs, containing 3383 and 2849 bp, which encode an identical open reading frame until amino acid 649 and then differ by the last few amino acids at the C-terminal (ASGASSA for CTL1a and LRKR for CTL1b) and by their 3′ untranslated region. Moreover, database analyses further indicated a similar 3′ splice variant for CTL5, while CTL2 shows alternate splicing for both the N- (data not shown) and C-termini (Fig. 1c). On the contrary, no indication for splicing of CTL3 and CTL4 was found. An alignment of the C-terminal protein sequences generated by these 3′ alternative spliced variants of the members of the CTL family is shown in Fig. 1(c).

Tissue distribution of choline transporter-like family members

DNA probes specifically recognizing the CTL1, CTL2, CTL3 and CTL4 transcripts from rodent were synthesized and hybridized with northern blots of rat mRNAs extracted from different regions of the nervous system or peripheral tissues (Fig. 2). The DNA probes were not further selective for any splice variant as they were derived from sequences located in the central coding region. CTL1 was widely expressed in all brain areas studied as well as in the spinal cord and sciatic nerve as a major 3.5-kb and a minor 5.0-kb transcript (Fig. 2a). Unlike CTL1, the other rat CTL family members were seen to be either poorly expressed, as was the case for CTL2 (Fig. 2b), or barely detectable, as seen for CTL3 (Fig. 2c) and CTL4 (data not shown), in the nervous system samples examined. CTL2 expression was easily observed as a 4.0-kb transcript in tongue, muscle, kidney, heart and lung but was fainter in testis, intestine and stomach (Fig. 2b). Despite evidence for splice variants of mouse CTL2 from the database survey, the single transcript observed suggests a potential difference in the relative frequency of splice variants between mouse and rat or an identical size for both variant transcripts. A single 2.5-kb transcript was observed in different tissues for CTL3 and CTL4 (Figs 2c and d and data not shown). CTL3 was expressed in a decreasing manner in the colon, kidney and ileum, while CTL4 was found at a high level in the intestine, kidney, stomach and, to a lesser extent, testis and lung. As no rodent CTL5 cDNA sequence of sufficient length and quality was available to synthesize a probe, CTL5 expression was tested by RT-PCR using primers whose sequence was deduced from conserved human exons found in the mouse and rat genomes. The CTL5 signal was faint in the rat brain and higher in the spinal cord but always much less abundant than the CTL1 signal (not shown).

Figure 2.

Tissue distribution of choline transporter-like (CTL) 1–4 transcripts. (a–d) Northern blot analysis of CTL1, CTL2, CTL3 and CTL4 transcripts in rat tissues. RNA blots contained 5 µg of poly(A)+ per lane in (a, b and d) (except for the sciatic nerve where 30 µg of total RNA was used) and 2 µg of poly(A)+ per lane in (c). They were hybridized with rat CTL1, mouse CTL2, CTL3 or CTL4 probes as indicated. Blots were exposed to X-ray films for 3–6 days (a, b, right panel and c) or 15 days (b, left panel and d). Effective RNA quantity of each lane was determined by reprobing the blots with a rat β-actin probe (bottom panels). Molecular weights (kb) are indicated on the left side.

Developmental expression of CTL1 in oligodendroglia cells and neurones in the nervous system

As CTL1 was the CTL family member most widely expressed in the nervous system, we decided to characterize its spatial and temporal expression in the brain using a cRNA probe recognizing both CTL1a and CTL1b splice variants. Initial ISH experiments have shown that, in adulthood, CTL1 is expressed in both neurones and oligodendrocytes (O'Regan et al. 2000). A similar profile of distribution was observed earlier, during post-natal development. In the hippocampus, a high staining intensity was persistently observed between post-natal day (P)5 and adulthood in the granule cells of the dentate gyrus and pyramidal cells of the Ammon's horn with a higher labelling of CA1 compared with CA2/CA3 (Figs 3a–e). Likewise, in the cerebellar cortex the granule cell precursors of the external granule cell layer at P5 and P13 and the mature granule cells of the internal granule cell layer between P5 and adulthood expressed CTL1 (Figs 3f–k). In contrast, the spatial and temporal expression of CTL1 transcripts was regulated in the fibre tracts with a distribution profile reminiscent of that for MBP (Ferry et al. 2000), an oligodendrocyte marker (Shiota et al. 1989). The signal increased between P5 and P13 and was predominantly observed in the deep white matter of the cerebellum and then in the white matter of the various lobes (Figs 3f–h). From P13 to P20, it was maintained in the fibre tracts of the cerebellum and then decreased until adulthood, when positive cells were still detected as small chains in the white matter, suggesting that they correspond to oligodendrocytes, or as scattered cells in the granular and molecular layers (Figs 3j and k). The density of CTL1 mRNA peaked later in the fibre tracts of the rostral part of the brain. This could be observed in the corpus callosum where the labelling still increased between P13 and P20 (Figs 3b–d) before it decreased in the adult (Fig. 3e) as well as in other fibre tracts, such as the lateral olfactory tract, fimbria, internal capsule or optic tract (Fig. 3l and data not shown). As a control for the specificity of hybridization, the sense CTL1 riboprobe gave no signal (Fig. 3m). Experiments using both MBP and CTL1 riboprobes confirmed that CTL1 was synthesized by oligodendrocytes, as shown in the adult brainstem where about 60–70% of MBP-positive cells expressed CTL1 (Figs 4a and c). Altogether, these data indicate that CTL1 expression occurred in oligodendrocytes during the period of myelination and in mature oligodendrocytes. Moreover, we have identified, by double-labelling ISH experiments, that more than 80% of choline acetyltransferase-positive cells in the motor nuclei as for instance the facial nucleus express CTL1 (Figs 4b and d). On the contrary, choline acetyltransferase and CTL1 were not colocalized in the medial septum, indicating that CTL1 is not expressed in all cholinergic neurones (data not shown). Moreover, the latter result constituted an internal control allowing validation of the step of inactivation of the alkaline phosphatase between the two successive steps of colorimetric detection required in double-labelling ISH experiments.

Figure 3.

Developmental expression of CTL1 in rat nervous system. In situ hybridization experiments were carried out on sagittal brain sections from post-natal day (P)5 (a and f), P13 (b, c, g and h), P20 (d, i, l and m) or adult (Ad) (e, j and k) or on sagittal dorsal root ganglion sections from adult (n–p) using CTL1 antisense (a–l, n and p) or sense (m and o) riboprobes. (c, h, k and p) Magnifications of (b, g, j and n), respectively. CA1, CA1 area of Ammon's horn; CA3, CA3 area of Ammon's horn; Cb, cerebellum; cc, corpus callosum; Cx, cortex; DG, dentate gyrus; e, external granule cell layer; f, fibre tract; fi, fimbria; g, granule cell layer; hip, hippocampus; i, internal granule cell layer; ic, internal capsule; lo, lateral olfactory tract; m, molecular layer; opt, optic tract; p, purkinje cell layer. Scale bars: 1 mm (l and m), 200 µm (a, b, d–g, i and j), 150 µm (n and o). The black arrowheads in (p) indicate the labelling of a large and a small sensitive neurone. The CTL1 sense riboprobe detected no signal as shown on a sagittal section from P20 rat (m) and from adult dorsal root ganglia (o).

Figure 4.

CTL1 colocalizes with oligodendrocytic and cholinergic markers. Double in situ hybridization experiments on frontal sections of adult rat brain used CTL1 (a and b), myelin basic protein (MBP) (c) or choline acetyltransferase (ChAT) (d), respectively. Arrowheads indicate cells coexpressing CTL1 and MBP in the brainstem (a and c) or CTL1 and ChAT in the facial nucleus (b and d) Scale bars: 25 µm (a and c), 50 µm (b and d).

In addition to its expression in the brain and spinal cord (O'Regan et al. 2000 and present data), we detected CTL1 mRNA in sensory neurones of the adult rat dorsal root ganglia (Figs 3n and p). Both large diameter sensitive neuronal cells, which mostly possess myelinated axons and respond principally to low threshold stimuli, and small diameter neurones, which have unmyelinated axons and are principally nociceptors and thermoceptors (Michael and Priestley 1999), were labelled. No labelling was observed with the sense riboprobe (Fig. 3o).

Analysis of CTL1 splice variant distribution

In order to further characterize the expression of CTL1 splice variants, we generated specific DNA and cRNA probes corresponding to the nucleotide sequence common to both splice variants (CTL1) or to the untranslated 3′ region of each of them (CLT1a and CTL1b). The CTL1 probe recognized both a 3.5-kb transcript mainly expressed in brain and a 5.0-kb transcript predominant in ileum and colon, while the CTL1a and CTL1b probes detected selectively the 5.0- and 3.5-kb transcripts, respectively (Fig. 5). This latter transcript was also detected in the rat glioblastoma cell line C6 in accordance with the size of the major transcript found in cerebral tissue but it was not observed in the mouse neuroblastoma cell line N18. A 4.0-kb transcript detected by the CTL1 probe and, to a lesser extent, the CTL1a probe in both cell lines may correspond to the CTL1c splice variant defined in human immune cells (Wille et al. 2001). This point was tested by RT-PCR showing that, in addition to CTL1a, the CTL1c splice variant is also present in rat brain and spinal cord as well as in glioma cells. However, the N18 cell line showed no detectable signal for CTL1c (not shown).

Figure 5.

Brain and peripheral tissue distribution of CTL1a and CTL1b splice variants. Northern blot analysis of RNA from adult rat brain, ileum or colon or from the C6 and N18 cell lines using probes corresponding to nucleotide sequences either common to both transcripts (CTL1) or specific for CTL1a and CTL1b. Poly(A)+RNA (1, 8 and 3 µg) was loaded for brain, ileum and colon, respectively. 25 µg of total RNA was used for the cell lines C6 and N18. Effective RNA quantity of each lane was determined by reprobing the blots with a rat β-actin probe (bottom panels). Blots were exposed to X-ray films for 6 days [choline transporter-like (CTL)] or 1 h (β-actin). Molecular weights (kb) are indicated on the left side.

Brain sagittal sections from young (P13) rats were then hybridized with cRNA-specific CTL1a and CTL1b probes. In the cerebellum, CTL1a (Figs 6a and c) and CTL1b (Figs 6d and f) were both strongly expressed in chains of cells in the fibre tracts indicating labelling of oligodendrocytes. However, CTL1a, but not CTL1b, was also found in the granule cells and/or their precursors within the internal and external granule cell layers (Figs 6c and f). Thus, CTL1b labelling appeared to be restricted to oligodendrocytes, which was consistent with its expression in fibre tracts in other brain areas. Likewise, in other areas such as the hippocampus, CTL1a expression was detected in neuronal cells (data not shown). In addition, this differential expression of CTL1 variants persisted in adulthood such as in the granule cells of the dentate gyrus or the CA1–CA3 pyramidal cell layer (data not shown).

Figure 6.

Overlapping but not identical distribution of CTL1 splice variants in the developing rat cerebellum. Specific antisense CTL1a (a and c) and CTL1b (d and f) riboprobes led to a different hybridization pattern on sagittal sections from post-natal day (P)13 rat cerebellum. Both CTL1a and CTL1b labelled fibre tracts (f), while only CTL1a was detected in the external (EGL) and internal (IGL) granule cell layers. (c and f) Higher magnifications of (a and d). Results obtained with CTL1a and CTL1b sense riboprobes are shown in (b and e), respectively. mol, molecular layer; PL, Purkinje cell layer. Scale bars: 500 µm (a, b, d and e), 100 µm (c and f).

Analysis of choline uptake by rat CTL1a and CTL1b splice variants

We had previously shown that torpedo CTL1 (tCTL1) increased high-affinity choline uptake in a choline transport-deficient yeast strain (O'Regan et al. 2000) as well as in Xenopus oocytes (O'Regan and Meunier 2003). We have now evaluated rat CTL1 (rCTL1) splice variants for their capacity to influence choline transport in a eukaryotic cell line. All of the cell lines which we have screened so far by RT-PCR express a CTL1 sequence common to both splice variant transcripts (S.O'R., personal observation). However, N18 cells expressed only low levels of a transcript recognized by this sequence (Fig. 5) and thus appeared to be the best choice for experiments aimed at addressing the question as to whether CTL1a and CTL1b have the same effect on choline transport. N18 cells were transiently transfected with plasmid constructs coding for CTL1a and CTL1b, both of which drove strong mRNA expression (Fig. 7c), and then evaluated for [3H]choline uptake. Both showed a similar modest increase in uptake compared with the basal level of choline uptake constitutively present in N18 cells transfected with the empty plasmid. This effect was due to significant (p < 0.05, n = 4) increases in apparent Vmax from 1.01 ± 0.08 pmol/mg protein in 4 min in controls to 1.65 ± 0.08 and 1.53 ± 0.12 pmol/mg protein in 4 min after expression of rCTL1a and rCTL1b, respectively. The Km values for the three conditions were very similar at 34.7 ± 4.1, 33.4 ± 2.5 and 33.0 ± 4.0 µm choline (Fig. 7a). Inhibition of choline transport by the choline analogue HC-3 was only moderate and similar for control and CTL1a- or CTL1b-expressing cells (Fig. 7b). In all cases, about 40% inhibition was obtained with 10 µm HC-3, suggesting that the Ki value was above this value. Hence, the native choline transporter in N18 cells and the CTL1a- and CTL1b-induced choline transport displayed similar affinities for choline and HC-3 but transfection with CTL1 increased the transport rate. A pure Vmax effect is often interpreted as a change in the number of transporters residing in the plasma membrane for solute transporters and further indicates that the general mechanism of transport is the same before and after expression of exogenous CTL1. A possible explanation for these results is that the endogenous form of CTL1 has the same choline transport characteristics and that the appearance of CTL1 proteins at the cell membrane is limited in some way beyond the transcriptional level. These results corroborate those obtained in yeast (O'Regan et al. 2000) and oocytes (O'Regan and Meunier 2003), where CTL1 expression was also linked to a modest gain in choline transport activity but each paradigm shows some specificity in substrate and inhibitor affinities. From this point of view, it is interesting to note that a decrease in the choline uptake affinity has been reported to occur when neuronal and glial cells were pre-loaded with choline (Wong et al. 1985), showing that cellular choline transport is sensitive to the internal levels of substrate and therefore presumably to the overall rate at which choline is metabolized by the cell.

Figure 7.

CTL1a and CTL1b splice variants increase choline uptake in transfected neuroblastoma N18 cells. (a) At 24 h after transfection with the control, CTL1a or CTL1b pcDNA3 constructs, mouse neuroblastoma N18 cells were tested for [3H]choline uptake. Cells accumulated more choline after transfection with the plasmids encoding CTL1a and CTL1b due to an increase in Vmax (p < 0.05, n = 4 for each condition) compared with control transfection with an empty plasmid. (b) In all cases, the addition of 10 µm hemicholinium-3 (HC-3) inhibited [3H]choline uptake by about half. Data are means ± SEM of four independent experiments. (c) A northern blot analysis of transfected cells with a probe corresponding to a common coding sequence for CTL1 showed that plasmids encoding CTL1a or CTL1b drove strong mRNA expression in N18 cells. Molecular weights (kb) are indicated on the left. Actin signals on the same blot are shown as loading controls.

Discussion

The present work shows that the family of choline transporter-like proteins (CTL family) is widely conserved among eukaryotes, evolving from a single gene in simple unicellular organisms such as yeast to five genes in mammals, with an additional level of complexity engendered by alternative splicing for several of these genes. Our study indicates that individual CTL family members are preferentially expressed in different tissues. Indeed, CTL2 and CTL4 have complementary patterns of expression in peripheral tissues, while CTL1 is the main member of this family to be expressed in the developing and mature rodent brain, suggesting that the corresponding proteins do not lead to redundant activities in a given tissue.

The high level of expression of CTL1 in neuronal and glial cells argues for the existence of a major functional role for CTL1 both during development and in the adult. The strong expression of CTL1 in oligodendrocytes, the myelin-synthesizing cells of the brain which require high levels of choline to synthesize phosphatidylcholine and sphingomyelin (Vos et al. 1997), suggests CTL1 involvement in the formation and/or maintenance of the myelin sheath. In agreement with this hypothesis, the distribution profile of CTL1 in the developing rat brain is correlated with the timing of active myelination starting around P10 and reaching a maximum at P18 (Baumann and Pham-Dinh 2001). Both the temporal and spatial regulation of CTL1 expression in the white matter parallels that of the myelin marker MBP. In addition, in the peripheral nervous system, CTL1 is highly expressed in the sciatic nerve. The strong expression of CTL1 in myelinating tissues appears to have been conserved throughout species evolution as an orthologous CTL1 protein has been detected in both oligodendrocytes and Schwann cells in the CNS and peripheral nervous system of the electric fish T. marmorata (Meunier and O'Regan 2002). Here, we further show that the two isoforms of rat CTL1 have an overlapping but not identical tissue distribution and cell-specific localization, CTL1a and CTL1b both being expressed in oligodendrocytes, while CTL1a is also expressed alone in neuronal cell populations.

The phenotypic diversity of neuronal cells expressing CTL1 indicates the probable importance of the protein for a process common to a wide variety of neurones including cholinergic neurones. In the granule cells of the cerebellar cortex and dentate gyrus, the synthesis of membranes might again be suggested as a potential role for CTL1. Indeed, these neurones are generated during the first post-natal weeks and probably need high choline levels for neurite elongation. Such a hypothesis is also in agreement with the delayed up-regulation of CTL1 expression recently reported in a model of facial motor nerve axotomy as being concomitant with the process of axonal regeneration (Che et al. 2002). Conversely, the expression of the high-affinity choline transporter CHT1, initially characterized as the transporter of cholinergic nerve terminals, is down-regulated in a way correlated with the loss of choline acetyltransferase immunoreactivity after transection of the hypoglossal nerve (Okuda et al. 2000; Misawa et al. 2001). Thus, the redundant expression of two distinct high-affinity choline transporters in motorneurones would allow choline uptake for a different fate within those cells.

With respect to the alternative splicing of CTL1a and CTL1b, the difference in the 3′ non-coding region might influence, for instance, the kinetics of RNA turnover (Curatola et al. 1995), while the alternative C-terminal coding sequence might control interactions with different sets of associated proteins or be involved in targeting of the protein to different subcellular compartments. Indeed, an RXR motif occurring in CTL1b has been classically identified as a potential endoplasmic reticulum retention/retrieval motif in ion channels or G protein-coupled receptors (Zerangue et al. 1999; Pagano et al. 2001). For example, the trafficking of the NMDA receptor between endoplasmic reticulum and the plasma membrane is controlled by post-synaptic density protein PSD-95/septate junction protein Discharge/tight junction protein ZO-1 (PDZ) (one of the most common modular protein interaction domains) and endoplasmic reticulum retention/retrieval motifs that are generated by alternative splicing (Scott et al. 2001). Whether this endoplasmic reticulum retention/retrieval motif is functional in CTL1b remains to be demonstrated, nevertheless its existence would be concordant with the identification of the site of phosphatidylcholine synthesis in the endoplasmic reticulum (Bishop and Bell 1988). However, such a functional contribution in vivo is speculative until antibodies are available.

Initially characterized as a suppressor in yeast with a choline transport mutation (ctr1), in addition to an as yet undefined conditional choline auxotrophic mutation (ise) (Matsushita and Nikawa 1995), CTL1 has been linked to an increase in choline uptake in N18 cells in the present report, as well as previously in Xenopus oocytes (O'Regan and Meunier 2003). The present work further shows that the two major CTL1 carboxy-terminus splice variants are endowed with similar stimulatory effects on choline uptake activity. It remains possible that the CTL proteins are not transporters themselves but are needed to activate, chaperone or enhance choline transport provided by endogenous pathways, particularly as the substrate and inhibitor affinities for choline transport were unchanged in the transfection study. Moreover, other aspects of CTL family protein function in yeasts (Zufferey et al. 2004) and vertebrates are also beginning to emerge, such as the observation that a CTL1 splice variant known as CDw92, recently described as a surface antigen in haematopoietic cells, plays a role in the negative regulation of the immune response in dendritic cells (Wille et al. 2001). More recent studies have shown that antibodies to CTL2 cause hearing loss associated with structural disruption when infused into the inner ear of guinea pigs where the protein is expressed on the surface of supporting cells surrounding the hair cells; this study further suggests that CTL2 immunoreactivity may play a pathological role in human autoimmune hearing loss (Nair et al. 2004). Both of these studies indicate a crucial role for CTL proteins in cellular function although neither specifies the nature of that role. Further studies will be necessary to clarify the degree of functional conservation between the different CTL members present in humans as well as their physiological and pathological impact.

Ancillary